Cannabis cross-price elasticities

It's been a bit of an open question whether legalised marijuana would lead to more or less alcohol use.  If the two goods are complements, say if people liked drinking while consuming cannabis, then any increase in cannabis use could yield greater alcohol use. If they were substitutes and people smoked instead of drinking, alcohol use could drop.

RAND surveys some of the more recent evidence. 

Different studies also examine different time periods, and the laws have been changing over time. Early state laws (such as the medical cannabis legislation California passed in 1996) tend to allow broader qualifying patient conditions, legal home cultivation and less oversight of dispensaries. Differences in policies may lead to different effects on cannabis use, and possibly alcohol use. And the laws' impact may evolve over time as the market expands or as federal enforcement shifts.

A recent working paper out of the University of Connecticut and Georgia State University has received a fair bit of attention as the latest in this series of attempts to shed light on the issue of whether alcohol and cannabis are substitutes or complements based on evidence from medical cannabis laws. The authors examined changes in alcohol sales at grocery and convenience stores and other outlets. They found that cannabis and alcohol are strong substitutes, with medical cannabis implementation being associated with a 15 percent reduction in monthly alcohol sales.

That is a surprisingly large effect, equivalent to what we would predict if the price of alcohol increased on the order of 30 percent. The effect seems especially large considering that during the study period of 2006 to 2015, the newer state medical cannabis programs that drive the main result were more restrictive and had low participation rates, typically involving less than 1 percent of the population (PDF). Of course, these medical laws could have effects that reach beyond the registered patient population if they made it easier and cheaper for non-patients to access cannabis, or if the laws caused the public to change its attitudes about cannabis and alcohol use more broadly. Much more needs to be learned about what's driving the results in this working paper.

They warn that these results on medical marijuana might not apply to recreational cannabis laws, as alcohol tax revenues in Colorado and Washington have remained fairly stable since marijuana shops there opened.